• 

  Fake News

• 

  Real News

• 

  Taking sample input

MAIS202 FALL 2020 FINAL PROJECT: Fake News Detector

This is the McGill MAIS202's Final Project. The goal in this project is to produce a "true" or "fake" classification on any news input. The proposed and implemented algorithm is the classical Naive Bayes Algorithm. The biggest challenge in the project is in the pre-processing of inputs, it turns out that human languages are rather "noisy" in a sense that there are a lot of redundancies in words, and grammatical rules commonly causes errors and are often useless in the algorithmic computation. Therefore I have implemented extensive natural language pre-processing, methods such as "stop-words removal" and "lemmatisation" were used to improve the classification accuracy. The BagOfWords algorithm is used to compute the relative probability of the Naive Bayes Algorithm due to its accuracy and efficiency. After Grid-searching on the Multinomial Algorithm and by implementing the best parameters, a test accuracy of 97% was achieved.

Built With

• flask
• html
• jupyter-notebook
• python